Septerra Core: Legacy of the Game Part 03: Stuck in a Loop
Septerra Core: Legacy of the Creator marked its 20th anniversary since publication late in 2019. What follows is a deep dive retrospective on the making of this cult classic RPG!
(BTW – In case you missed them, here’s Part 01 or Part 02)
In 1996 Septerra Core was a fledgling project at the Chicago offices of Viacom New Media (VNM). We had begged and borrowed time and resources from other teams to sell the project to the suits. Now, pre-production was ready to start in earnest. But while the team was raring to begin full development, we soon found a seemingly unending array of hurdles springing up underfoot.
The Team Grows
The core team (No pun intended) in ’96 would grow to (from left to right) Cathi Court (Producer), Roger D (Tools Programming), Brian Babendererde (Game Design), Kathy Tootelian (front, Lead Artist), Alisa Kober (back, Art Production), Mark Manyen (PS01 Engine Lead), and Tom J (Windows 95 Engine Lead, not pictured).
Up to this point, the team had consisted of myself, lead artist Kathy Tootelian, and producer Cathi Court. It was time to grow.
VNM had already been looking to hire artists for several projects and I found myself invited to check out reels and resumes along with the lead artists from multiple teams.
So. Many. Bland. Reels.
But, there was one that stood out. While most contained fairly typical stuff, this one was chock full of 3D models based on anime characters. Priss from Bubblegum Crisis busting out of her hardshell is the one that I distinctly remember. No one else in the room that day had any idea who that was—but I knew this artist would have a passion for the same things I did.
That artist was Kye, a young man with a background in architecture, but no game development experience.
“During the (job) interview BMAN showed me a couple of his character concept illustrations,” Kye recalls. “…I was sold and begged for the job.”
In fact, Kye took it upon himself to model a version of Lobo from my concept drawing in his spare time. He clearly understood what we were building and we hired him soon after.
Select any image enlarge, and select again to learn more.
Kye did these before he was hired. I don’t remember them being a test—it’s my recollection that he did them on his own initiative, based off a drawing I shared with him. Scanned from a print out.
Kye did these before he was hired. I don’t remember them being a test—it’s my recollection that he did them on his own initiative, based off a drawing I shared with him. Scanned from a print out.
This was a lower-poly version of Lobo he did at the same time, showing how the character might work in a real-time game engine. Scanned from a print out.
This was a lower-poly version of Lobo he did at the same time, showing how the character might work in a real-time game engine. Scanned from a print out.
Kye also did a “Super Deformed” version before he was hired. I thought this was really cute. However, when we hired him I think he was surprised to learn that we were going to eschew the “Chibi” style in Septerra Core. Scanned from a print out.
Kye also did a “Super Deformed” version before he was hired. I thought this was really cute. However, when we hired him I think he was surprised to learn that we were going to eschew the “Chibi” style in Septerra Core. Scanned from a print out.
This was also Kye, based off a drawing of Led I had been working on. I believe this was also done on his own initiative around the same time as we hired him (This is not the model that would be used in the game). Scanned from a print out.
This was also Kye, based off a drawing of Led I had been working on. I believe this was also done on his own initiative around the same time as we hired him (This is not the model that would be used in the game). Scanned from a print out.
One of Kye’s early and important contributions was taking over on concept drawings—designing enemies, creatures, ships, mecha, weapons, and non-player characters (NPCs). Kye once told me that we basically paid him to work on his drawing skills for something like six months, because for quite a while that was his main job. Later artists would add their own equally incredible designs, and I even found the time to contribute one or two battleship designs (mostly because I still longed to contribute more concepts of my own), but by and large the lion’s share of the initial concept sketches for Septerra Core were drafted by Kye. While I had set the tone for the games’s visuals with the main character designs, it was Kye who would ultimately define much of the style in those early days.
Another developer that came on early was Mary K Omelina, an artist from an outside studio that had helped out on animation work for Beavis and Butt-head in Virtual Stupidity. She joined VNM and worked with us to develop possible looks for the final game animation. She soon got a crash course in anime, as I brought in my personal collection to show as examples of the style were going for. It was all quite different than the world of American animation she was used to…
“I finally understood many years later what anime running was” – Mary K.
Kurt Mitchell was another artist that pitched in early to help visualize the world. Kurt passed away just a few weeks ago (July 2020), so I wanted to give a special call out in his memory.
Kurt and I used to hang out at Viacom New Media and talk about fantasy art and illustration. As an older artist, Kurt had been around the block, and I learned a lot from him. When he found out I was working on a “unique fantasy world,” he kinda scoffed, assuming that this creation would just be more of the same medieval swords and elves! Old Kurt the curmudgeon gave me examples of truly unique worlds, going through a rundown of Philip Jose Farmer’s excellent Riverworld series as well as Niven’s Ringworld. I then began to tell him how Setperra worked, from Shell 01 down to the Core, the spine, the conjunction, and how all the cultures interacted. Surprised, he absolutely lit up!
Kurt got excited and wanted to help visualize this world, which we absolutely needed help with—Septerra was unique, and we needed to figure out how its skies would look. This image was one of his early attempts.
Early visualization of what a Septerra vista might look like—background element by the late, great Kurt Mitchell. Kurt would go on to do some of our side-view, cinematic backgrounds as well as contributing early elements to our isometric levels.
I liked it so much that it made it all the way to the final box art for the North American release!
Kurt was excited to talk about the making of the game before he passed. When I asked for any stories from the old VNM days, this is what he had to say:
“The time we had a power outage in the (VNM) building and it was pitch black. I led some people out of the building using the glowing eyes of my 16 inch Batman action figure.” — Kurt Mitchell
We’ll miss you, Kurt!
A few more full-time 3D modelers came on to build characters and environments, while a couple programmers joined us to begin building the technical foundation for the game engines.
Wait, BMAN, did you say, engines, as in, plural? Yep, believe it or not, Septerra Core was originally slated for development on two platforms: the PC and the Playstation 1. Tom J was working on the game engine for the PC version while Mark Manyen was working on the PS1. Roger D was another programmer hired to develop the many tools we would need to layout environments, script logic, and orchestrate scenes and dialogue.
“The first time in BMAN’s office completely blew my mind. For the first time, I was not the biggest geek in the room.” – Mark Manyen on the absurd amount of toys I had in my office.
Brainstorming with the newly assembled team of art and programming became a daily occurrence. We gathered around the conference table in the Septerra Suite, where I would lead everyone through the concepts that were bubbling up from my admittedly warped little brain. Everyone was invited to contribute ideas to the world, story, or gameplay, and every day Septerra came more and more into focus as we developed cultures and started illustrating the look of each world shell. Concepts were tossed around, tossed out, or replaced. The back and forth flow of ideas and the team’s feedback helped me shape the world and the story into a more concrete outline as I moved towards writing a first outline of the script.
Newspaper clipping from later in the year, when the Chicago Tribune did a story on Viacom New Media (later re-named Rabid Entertainment, but we’ll get to that next time…) This was the Septerra suite conference room, and a growing wall of character, creature, and location designs were tacked up on every available space. The characters are pinned up behind us, but the location designs would have been on the opposite wall.)
Resolution Revolution
While previous games had been rendered at a screen resolution of 320×240 pixels, Septerra Core was aiming for the monstrously huge (at the time anyway) resolution of 640×480. It can be difficult to impress the importance of this sea change in resolution onto a modern reader—640×480 was double the resolution, which meant four times the number of pixels on screen and four times the memory required. Everyone was excited to move into this new world, as more pixels equaled higher fidelity graphics, more detail, and less of the “blocky” look of the previous generation of games.
Unfortunately, the world of sprite animation was not quite ready for this brave new world. All the well-made tools available to us at the time had been created with the previous generation of graphics in mind, geared at creating sprites at 320×240, with artists often placing one pixel at a time to create each frame (something we called “pixel pushing.”) But with the resolution doubled, those sprite tools and techniques were suddenly no longer very feasible.
No Big Heads Here
Our first alternative was doing the characters in real-time 3D, overlaid onto 2D backgrounds. This is the approach that Final Fantasy 7 would eventually take a few years later, but we abandoned it fairly quickly. One thing I wanted to avoid in SepterraCore was the JRPG phenomenon known as “Super Deformed” characters. You’ve seen it before, the “big head/small body” look that many Japanese games use when representing their characters. It was a useful tool in lower-resolution games—if an entire party of characters and enemies had to be on screen at once then the sprites needed to be small, so making them exaggerated and cartoony made them easier to render and identify. But I desperately wanted to avoid this “cute” look, aiming instead for something dramatic— not necessarily “realistic” so much as “Western graphic novel meets Eastern anime.” Because of this, our early 3D tests were not promising, with characters that had to be reduced down to simple polygons and textures, losing much of their charm and that “cool factor.”
The early, “dirt-cheap” 3D tests. Select any image enlarge, and select again to learn more.
A simple 3D mock up of Maya done as a test for an in-game 3D engine at Viacom New Media. The artist did these on his own time, and noted: “The model and the textures SUCK, but it gives you an idea what you can do.”
A simple 3D mock up of Maya done as a test for an in-game 3D engine at Viacom New Media. The artist did these on his own time, and noted: “The model and the textures SUCK, but it gives you an idea what you can do.”
Another shot of the Maya real-time 3D model test. These were intended to prove that we could have “3 main characters and 5 enemies and still have 1000 polygons of scenery (running at 15 frames per second).
Another shot of the Maya real-time 3D model test. These were intended to prove that we could have “3 main characters and 5 enemies and still have 1000 polygons of scenery (running at 15 frames per second).
This was a quick and dirty test with no polishing or revisions, but it was clear that unless we went with more cartoony proportions we were going to have a hard time making our detailed characters look cool. The only versions of this real-time 3D test are scanned from old grayscale print-outs.
This was a quick and dirty test with no polishing or revisions, but it was clear that unless we went with more cartoony proportions we were going to have a hard time making our detailed characters look cool. The only versions of this real-time 3D test are scanned from old grayscale print-outs.
Research and Development
Another alternative was to create traditional 2D animation. Digital ink and paint was something that was just starting to appear in film and TV animation—instead of painting images onto celluloid to create each frame of animation, programs like Toon Boom were now allowing the process to be done on a computer as digital ink and paint. Rather than create the characters laboriously one pixel at a time, we could create them as hi-res, traditional animation, which could then be used as game sprites and for cinematic story sequences.
Mark K had a background in traditional animation, so she did some test poses of both Maya and Corgan.
Select each image to see the never before seen traditional animation test drawings! Select again to learn more about each one.
Maya traditional animation test. Mary K was working to break the character down into something that could be animated.
Maya traditional animation test. Mary K was working to break the character down into something that could be animated.
Maya traditional animation test turn-around. Because I had only drawn Maya from the front, Mary K had to work with me to figure out how the costume would look from other points of view.
Maya traditional animation test turn-around. Because I had only drawn Maya from the front, Mary K had to work with me to figure out how the costume would look from other points of view.
Maya traditional animation turn-around. Mary K figuring out how Maya’s leg armor and hair would look from behind.
Maya traditional animation turn-around. Mary K figuring out how Maya’s leg armor and hair would look from behind.
Maya traditional animation test faces and expressions. Mary K had a background in traditional animation, and she did a good job converting that to a hybrid “east meets west” style.
Maya traditional animation test faces and expressions. Mary K had a background in traditional animation, and she did a good job converting that to a hybrid “east meets west” style.
Corgan traditional animation test, front view. Mary K did a great job here of figuring out the shape and feel of the armor.
Corgan traditional animation test, front view. Mary K did a great job here of figuring out the shape and feel of the armor.
Corgan traditional animation test turn-around. As with Maya, Mary K had to work out what Corgan’s costume looked like from other angles, as well as add a scabbard.
Corgan traditional animation test turn-around. As with Maya, Mary K had to work out what Corgan’s costume looked like from other angles, as well as add a scabbard.
Corgan traditional animation test: faces and expressions. These were some of my favortites, as Mary K did an excellent job interpreting Corgan’s facial features and expressions.
Corgan traditional animation test: faces and expressions. These were some of my favortites, as Mary K did an excellent job interpreting Corgan’s facial features and expressions.
While that might work for flesh and blood characters, hand drawing the robots and sophisticated hardware like ships and mecha was going to be more problematic. But what if we combined the two?
We did a test of Grubb riding on Runner’s back. Grubb was animated in Toon Boom while runner was rendered in 3D. It was an exciting test to be sure, but it became apparent that using this approach was going to be problematic in a couple ways.
Select each image to see the never before seen traditional 2D animation sprite tests! Select again to learn more!
The only known image of the fully animated 2D and 3D runner test. This test was done in color, but I was unable to find any moving footage. The only record of the test now seems to be this black and white still image rescued from an old design document.
The only known image of the fully animated 2D and 3D runner test. This test was done in color, but I was unable to find any moving footage. The only record of the test now seems to be this black and white still image rescued from an old design document.
An early mock-up of Septerra Core, as we worked out where user interface elements might go. However, the early 2D and 3D tests of Corgan are shown in the bottom right. Unfortunately, this test now only exists as a low-res, black and white image rescued from an early design document.
An early mock-up of Septerra Core, as we worked out where user interface elements might go. However, the early 2D and 3D tests of Corgan are shown in the bottom right. Unfortunately, this test now only exists as a low-res, black and white image rescued from an early design document.
Close up of the corgan sprite test. On top are 2D traditional animation versions, where the character was inked and painted in Toon Boom, then shrunk down to sprite size. The Corgan below is a 3D test render of an early model.
Close up of the corgan sprite test. On top are 2D traditional animation versions, where the character was inked and painted in Toon Boom, then shrunk down to sprite size. The Corgan below is a 3D test render of an early model.
First, we would have to animate the human characters several times over to cover all the different isometric viewpoints required for the game—talk about labor intensive! Second, we recognized that we wouldn’t have the manpower, experience, or budget to create fully fledged “anime quality” cinematic scenes for our story sequences. After all, most of our in-house animators were more versed in sprites and 3D than traditional cell animation.
But, the art director for VNM really pushed us to use Toon Boom and 2D drawings. When it came to the cinematic scenes he actually agreed they would be cost prohibitive to draw and animate by hand, but argued that we should instead simply cut these story sequences from the game—problem solved! I fought to keep them. I knew we were going to need cinematics to help sell the uniqueness of our world, to periodically take the player out of the three-quarter isometric view of the game, at least for the big, earth shattering battles and events. Otherwise, we’d just end up having characters pointing off screen and yelling, “Oh my God! Look over there, just off screen, that earth-shattering battle in the skies! How exciting! If only the player could see it!”
So, to hedge our bets, we also did tests building the human characters in 3D, even though we weren’t entirely sure that 3D software in the mid 90’s was going to be up to the task. We wanted our characters to be organic and avoid any sort of unnatural “polygonal” look. Happily, our test models proved this route was feasible. They were highly complex, built with curves instead of polys. There was no 3D game engine that could render them in real time, but we didn’t need that. Instead, the idea was to build the characters as detailed as we could, then pre-render them out in any view needed as 2D sprites. This meant we could create each model once, then use them for in-game sprites, dialogue portraits, and even cinematics.
Building the characters in 3D was the approach that eventually won out, and we began to create our characters and figure out the style of the models and renders.
Before Kye modeled the main characters, he did “turn-around” drawings to help define the proportions and better understand the costumes. Select any of these never before seen images to enlarge, and select again to learn more.
Kye’s turn-around of Maya. This was probably the first character he worked on bringing to life in 3D.
Kye’s turn-around of Maya. This was probably the first character he worked on bringing to life in 3D.
Kye’s turn-around of Grubb. I really like how he captured his personality!
Kye’s turn-around of Grubb. I really like how he captured his personality!
Corgan drawings by Kye. You can see he’s breaking down how the knee armor actually might work (because I certainly hadn’t given it enough thought when I drew it!)
Corgan drawings by Kye. You can see he’s breaking down how the knee armor actually might work (because I certainly hadn’t given it enough thought when I drew it!)
Kye’s rendering of Led—He had me check drawings like these with an eye for character proportions before modeling them in 3D.
Kye’s rendering of Led—He had me check drawings like these with an eye for character proportions before modeling them in 3D.
Kye’s Selina turn-around. The Chosen armor I had designed definitely needed some thought put into it if the character was going to raise her arms and swing a sword.
Kye’s Selina turn-around. The Chosen armor I had designed definitely needed some thought put into it if the character was going to raise her arms and swing a sword.
Kye’s drawing of Lobo. His background in architectural rendering meant Kye was excellent at drawing mechanical forms from multiple views on paper before modeling them.
Kye’s drawing of Lobo. His background in architectural rendering meant Kye was excellent at drawing mechanical forms from multiple views on paper before modeling them.
Kye’s version of Araym. I’m not sure why there is no side view. It’s possible this scan was made before a side view had been completed.
Kye’s version of Araym. I’m not sure why there is no side view. It’s possible this scan was made before a side view had been completed.
Kye working out Badu’s front and side view. Fun fact: Kye was the voice of Badu in the game—we recorded him saying phrases in Korean, then played it backwards with additional effects!
Kye working out Badu’s front and side view. Fun fact: Kye was the voice of Badu in the game—we recorded him saying phrases in Korean, then played it backwards with additional effects!
The characters were then modeled in 3D using complex (for the time) curves. Select any of these early character and enemy renders to enlarge, and select again to learn more.
Early test render of Maya. The character models were created with the knowledge that they needed to look their best when rendered out as smaller sprites. So details of hair and textures were kept fairly simple.
Early test render of Maya. The character models were created with the knowledge that they needed to look their best when rendered out as smaller sprites. So details of hair and textures were kept fairly simple.
Early test render of Grubb with effects and reflections on the goggles.
Early test render of Grubb with effects and reflections on the goggles.
Runner test render. Runner was changed to have an upright walk, which added a lot of human personality to his character.
Runner test render. Runner was changed to have an upright walk, which added a lot of human personality to his character.
Early render of Corgan. Shots like these were created to help sell the game, showcasing the characters.
Early render of Corgan. Shots like these were created to help sell the game, showcasing the characters.
Early test render of Led. Led’s pony tails were never quite animated the way we wanted in the olden days before cloth and hair physics.
Early test render of Led. Led’s pony tails were never quite animated the way we wanted in the olden days before cloth and hair physics.
Early test render of Lobo with additional lighting added.
Early test render of Lobo with additional lighting added.
Early Araym test render. This character probably had more moving parts that most, as each arm needed to open and fire.
Early Araym test render. This character probably had more moving parts that most, as each arm needed to open and fire.
Early render of Selina. Wherever possible textures were used to add details, especially for characters that would never be seen in an interstitial movie. The ridges on her torso armor are an example.
Early render of Selina. Wherever possible textures were used to add details, especially for characters that would never be seen in an interstitial movie. The ridges on her torso armor are an example.
Early render of Badu (unfinished model). Badu was one of the last characters designed, and was probably not fully modeled when these renders were made.
Early render of Badu (unfinished model). Badu was one of the last characters designed, and was probably not fully modeled when these renders were made.
Princess Alisa, named after Alisa Kober. Alisa was instrumental in helping me come up with the initial layered-world idea of Septerra, and would become the Art Director of the project at Valkyrie Studios.
Princess Alisa, named after Alisa Kober. Alisa was instrumental in helping me come up with the initial layered-world idea of Septerra, and would become the Art Director of the project at Valkyrie Studios.
Render of the Chosen soldier. Some simple textures and bump-mapping added interest to armor on many of the characters.
Render of the Chosen soldier. Some simple textures and bump-mapping added interest to armor on many of the characters.
Render of the Chosen Shadow Soldier. Simple effects were used to add textures to surfaces such as armor. These would reduce down to interesting grain when rendered at game size.
Render of the Chosen Shadow Soldier. Simple effects were used to add textures to surfaces such as armor. These would reduce down to interesting grain when rendered at game size.
Lens Beetle render, with lighting showing the rough textures of the carapace.
Lens Beetle render, with lighting showing the rough textures of the carapace.
Early render of theWorkbots (Gunbots). I’m not sure if we eventually added textures to the armor and gun metal, or rendered them clean like this to make their colors pop.
Early render of theWorkbots (Gunbots). I’m not sure if we eventually added textures to the armor and gun metal, or rendered them clean like this to make their colors pop.
Thunder Lion render. Some models had much “dirtier” textures, especially the exoskeletons of creatures and insects.
Thunder Lion render. Some models had much “dirtier” textures, especially the exoskeletons of creatures and insects.
Toon Boom Goes Boom
Even though we now had a way to make the characters, we would need a new tool to sequence the raw renders into the sprite animation sequences that could be used in a game engine. While much of the American gaming industry was using software like Deluxe Paint, we had long been spoiled by Japanese and in-house tools that had been built to put more power over frame sequencing and game play interaction into the hands of artists. Unfortunately, those tools had been built for the previous generation of lower resolution games and would need to be overhauled, and that was going to take time and money.
Once again, the art director suggested using Toon Boom. (Did he own stock in the company?) But, his point was valid: Toon Boom already sequenced animation—what if we could work with their engineers to modify it into something that could spit out animation the way a game needed it, complete with data such as collision boxes and center points (the information that tells the game engine how a character interacts with the world and other characters or effects), as well as complex timeline editing tools that allow artists to sequence and combine frames in ways wholly unique to games. We spent several months working with Toon Boom’s developers, even sending Mary K off to their offices in Montreal.
For a while this seemed like it would bear fruit. However, after several rounds of back and forth, it suddenly fell apart. I distinctly remember walking away from a conference call with Toon Boom where it had become clear that, though they thought they were working towards what we needed, and doing so in good faith, they just weren’t understanding the unique needs of game animation.
It was a disaster for us, as we were set back to square one: we would definitely need to build an in-house tool as a solution, a tool we could have spent the previous months getting started on.
(Note that In the long run, Toon Boom would eventually go on to build robust gaming tools, though I don’t think that happened until the mid 2010’s…)
Some of the early sprite tests and guides. Select any image enlarge, and select again to learn more about how sprite characters work.
An early proof-of-concept test render of Grubb in sprite form.
An early proof-of-concept test render of Grubb in sprite form.
A sprite guide sheet, showing how melee based characters would need 4 directions. Ever notice that Grubb changes which hand he’s using to hold his staff? That’s because we needed to save storage memory on disc (and in RAM), so the left-facing versions of the characters are mirror flipped and reused for the right-facing versions.
A sprite guide sheet, showing how melee based characters would need 4 directions. Ever notice that Grubb changes which hand he’s using to hold his staff? That’s because we needed to save storage memory on disc (and in RAM), so the left-facing versions of the characters are mirror flipped and reused for the right-facing versions.
Another guide sheet showing how artists needed to render out the sprites for ranged-based characters. Although still flipped, characters wielding guns needed double the positions/frames so that they could shoot in all directions.
Another guide sheet showing how artists needed to render out the sprites for ranged-based characters. Although still flipped, characters wielding guns needed double the positions/frames so that they could shoot in all directions.
Another sprite guide, this time showing how characters would turn to target enemies. To avoid the feet of the characters sliding, we intended them to turn their upper bodies to aim.
Another sprite guide, this time showing how characters would turn to target enemies. To avoid the feet of the characters sliding, we intended them to turn their upper bodies to aim.
This sprite guide shows how reference points are used to align characters and other objects in the world. The artist sets up both Maya’s sprite and the muzzle flash sprite with reference points. Maya’s reference point is a “center point,” and is used to show not only the center of the character, but also approximately where her character intersects with the ground / exists in the 3D space of an isometric environment. If the game engine lines up Maya’s center point with the muzzle flash’s reference point, then the muzzle flash will appear where it is supposed to, at the barrel of Maya’s rifle. Note that the reference points are not rendered in the game, so the player never sees them.
This sprite guide shows how reference points are used to align characters and other objects in the world. The artist sets up both Maya’s sprite and the muzzle flash sprite with reference points. Maya’s reference point is a “center point,” and is used to show not only the center of the character, but also approximately where her character intersects with the ground / exists in the 3D space of an isometric environment. If the game engine lines up Maya’s center point with the muzzle flash’s reference point, then the muzzle flash will appear where it is supposed to, at the barrel of Maya’s rifle. Note that the reference points are not rendered in the game, so the player never sees them.
The entire player character cast rendered out as early test sprites—Araym, Selina, Corgan, Grubb, Maya, Runner, Led, Lobo, and Badu. Enlarged fron the original 640×480 render for better visibility on modern monitors.
The entire player character cast rendered out as early test sprites—Araym, Selina, Corgan, Grubb, Maya, Runner, Led, Lobo, and Badu. Enlarged fron the original 640×480 render for better visibility on modern monitors.
Let’s Light this Candle Already!
Our small team had grown, but we wanted that team to get much bigger. Like, being green-lit bigger. VNM did allow us to ramp up, adding a few more artists to the team, including Dave P, Terry S, Matt L, and Mike C. We also “stole” a few people from the testing department. Testers Robb S, Gus Peterson, and Dave C started working in the suite with us, three enthusiastic guys looking to work their way into the development side of VNM. They quickly learned the tools we were building, taking content from the artists and getting it into our early game engines while adding their own creativity to the mix.
The team from later in ’96. From left: (standing) Gus Peterson, Matt L, Joe H, Kathy Tootelian, Roger D, Robb S, Mike C, Dave P, Terry S, Kye, Kurt Mitchell. (Sitting row) Cathi Court, Mark Manyen, Mary K Omelina, Alisa Kober, Brian Babendererde, Dave C.
More team shenanigans. Clockwise from top: Matt (hiding), Robb S, Cathi Court, Mark Manyen, Roger D, Mary K Omelina, Alisa Kober, Gus Peterson, Terry S, Kurt Mitchell, Kye, Dave P, Kathy Tootelian, Mike C, Dave C (and Brian Babendererde crowd-surfing).
But the suits wouldn’t actually green light the game. Instead, they just kept checking in with our progress and then funding more pre-production development.
We happily did this for a while, exploring and fleshing our concepts out, honing our story, our characters, and the look of the game. The Septerra Suite became riddled with storyboards, as concept drawings for each world shell went up on the wall—the top shell of the Chosen in its own row at the top, the row below that for the Junkers of shell 2, etc—all the way down to the “core at the floor.”
“I had just started on a project at Viacom New Media and the teams were spread out in individual little fort studios that had glass fronts. Every day I’d walk by the studio that had the concept pics of Septerra Core taped to the glass and I would let anyone that would listen know I wanted to get on that team. It was an artist’s dream to work on it. – Dave P.
A sample of some of the original concept sheets that would have been hung on the walls. These are all from the pencil of Kye, and many have never been seen before! Select any image to enlarge, and select again to learn more.
Doskias’ flagship. The Chosen battleships were designed to look as if they were grown, possibly from modified animals of the upper shell (such as Helgak). Kye also designed them to look like angels with outstretched wings when seen from the ground.
Doskias’ flagship. The Chosen battleships were designed to look as if they were grown, possibly from modified animals of the upper shell (such as Helgak). Kye also designed them to look like angels with outstretched wings when seen from the ground.
The original design for the Chosen Emperor Aspertine. When it came time to model this character, who is only seen a few times in-game, we didn’t have the time to bring this concept fully to life, simplifying it in the interest of time—something that I always regretted.
The original design for the Chosen Emperor Aspertine. When it came time to model this character, who is only seen a few times in-game, we didn’t have the time to bring this concept fully to life, simplifying it in the interest of time—something that I always regretted.
Concept of the Chosen throne room on Shell 01. This might have been done when we thought that the opening scene in the throne room was going to be a cinematic movie.
Concept of the Chosen throne room on Shell 01. This might have been done when we thought that the opening scene in the throne room was going to be a cinematic movie.
I had designed the first Chosen with Selina. Kathy added to that with her design for Doskias. Kye then used this foundation to explore and extrapolate all the different Chosen characters. Many of these were my favorite concepts for the game, and it was fun to see this culture of Chosen Lords and Ladies evolve and come to life.
I had designed the first Chosen with Selina. Kathy added to that with her design for Doskias. Kye then used this foundation to explore and extrapolate all the different Chosen characters. Many of these were my favorite concepts for the game, and it was fun to see this culture of Chosen Lords and Ladies evolve and come to life.
Kassandra was a character that would eventually be cut. In the game Doskias had once taken Selina as a lover, but cast this low-born woman aside as his ambitions to become the true descendant of Marduk grew. However, in the original script he actually left Selina for Lady Kassandra, a woman of noble-blood who’s political ties could help him in the court of the Emperor. Kassandra would have died during the confrontation on the lens platform. In the game, Selina believes Doskias can still be saved, and tries to stop him from using the lenses to bring light to the Core early. In the original script, it would have been Kassandra who shields Doskias from Selina’s attack, dying in the effort—this tragic act enrages Doskias and cements a hatred for his former lover, Selina.
Kassandra was a character that would eventually be cut. In the game Doskias had once taken Selina as a lover, but cast this low-born woman aside as his ambitions to become the true descendant of Marduk grew. However, in the original script he actually left Selina for Lady Kassandra, a woman of noble-blood who’s political ties could help him in the court of the Emperor. Kassandra would have died during the confrontation on the lens platform. In the game, Selina believes Doskias can still be saved, and tries to stop him from using the lenses to bring light to the Core early. In the original script, it would have been Kassandra who shields Doskias from Selina’s attack, dying in the effort—this tragic act enrages Doskias and cements a hatred for his former lover, Selina.
Princess Alisa concept. Named after the Alisa Kober, who was instrumental in helping to develop the early concept of Septerra’s multi-tiered world! I loved the fact that she alone had a unique color of armor (red, the color of the Imperial House), making her stand out as from the other Chosen.
Princess Alisa concept. Named after the Alisa Kober, who was instrumental in helping to develop the early concept of Septerra’s multi-tiered world! I loved the fact that she alone had a unique color of armor (red, the color of the Imperial House), making her stand out as from the other Chosen.
Unused concept of a two legged walking tank for Shell 01. It would have been driven by a Chosen soldier.
Unused concept of a two legged walking tank for Shell 01. It would have been driven by a Chosen soldier.
Chosen soldier. I asked Kye to come up with a bio-organic offspring of an Imperial Stormtrooper and an EVA from Evangelion.
Chosen soldier. I asked Kye to come up with a bio-organic offspring of an Imperial Stormtrooper and an EVA from Evangelion.
Concept for the Chosen Shadow Soldier, otherwise known as the “ninja” around the office. I believe Kye intended the apparatus on his back to generate a cloaking device. We loved the idea that the Chosen just kept bio-sculpting different soldiers for all the needs of warfare, from front line troops to heavy grenadiers, large tank-like forms, and these ninja strike teams. The question is whether their people wanted to be bio-engineered into barely human killing machines for the cause, or were they conscripts?
Concept for the Chosen Shadow Soldier, otherwise known as the “ninja” around the office. I believe Kye intended the apparatus on his back to generate a cloaking device. We loved the idea that the Chosen just kept bio-sculpting different soldiers for all the needs of warfare, from front line troops to heavy grenadiers, large tank-like forms, and these ninja strike teams. The question is whether their people wanted to be bio-engineered into barely human killing machines for the cause, or were they conscripts?
Concept for the pumping stations on Shell 02. The original idea was that these were built in the past when the Junkers were more advanced, with the help of the Watchers. They might have also been involved in Wind City’s export of water from Shell 03—while I always imagined that that endevour involved large water tank gondolas slung under living dirigible helgaks, we did discuss the idea of the pumping stations being used to pump water up to Shell 02. Pipes would need to telescope down to Shell 03 and periodically dip into large bodies of water whenever the rotation of the shells lined up.
Concept for the pumping stations on Shell 02. The original idea was that these were built in the past when the Junkers were more advanced, with the help of the Watchers. They might have also been involved in Wind City’s export of water from Shell 03—while I always imagined that that endevour involved large water tank gondolas slung under living dirigible helgaks, we did discuss the idea of the pumping stations being used to pump water up to Shell 02. Pipes would need to telescope down to Shell 03 and periodically dip into large bodies of water whenever the rotation of the shells lined up.
Concept for Tori. This was probably the first Junker designed after Maya.
Concept for Tori. This was probably the first Junker designed after Maya.
Concept for the Watchers. Both Kye and myself were heavily influenced by Mamoru Nagano’s Five Star Stories design aesthetic, and that manifested most prominently in the Watchers.
Concept for the Watchers. Both Kye and myself were heavily influenced by Mamoru Nagano’s Five Star Stories design aesthetic, and that manifested most prominently in the Watchers.
An early design for the helgak that would take Maya, Grubb, and Runner to Shell 03 in a movie sequence. I don’t think this was ever used in-game, and we went with something else.
An early design for the helgak that would take Maya, Grubb, and Runner to Shell 03 in a movie sequence. I don’t think this was ever used in-game, and we went with something else.
The flying helgak creatures of Septerra were always something that I wanted to include more of in the game. We had more designs that we ended up using, but I think they gave the world a unique flavor.
The flying helgak creatures of Septerra were always something that I wanted to include more of in the game. We had more designs that we ended up using, but I think they gave the world a unique flavor.
Military HQ for World Shell 05. I believe we used this for the Ankaran building. Shell 05 was intended to feel more science fiction based, and the designs were full of metallic, futuristic looking, military-industrial structures.
Military HQ for World Shell 05. I believe we used this for the Ankaran building. Shell 05 was intended to feel more science fiction based, and the designs were full of metallic, futuristic looking, military-industrial structures.
Wind City’s Bowman. Kye had to extrapolate details from my early design of Corgan, and then envision how the other soldiers and officers off Wind City would be armored and armed. Bowman is an officer, denoted by his more elaborate shoulder pads and sash.
Wind City’s Bowman. Kye had to extrapolate details from my early design of Corgan, and then envision how the other soldiers and officers off Wind City would be armored and armed. Bowman is an officer, denoted by his more elaborate shoulder pads and sash.
Concept for the Wind City cathedral. Kye’s unique use of “wind sails” in the buildings for this city gave it a unique style.
Concept for the Wind City cathedral. Kye’s unique use of “wind sails” in the buildings for this city gave it a unique style.
Concept for the undead beast. This enemy would become the mini-boss Drax in the graveyard / catacomb portion of the story, but was also intended to be used as a regular undead creature as well.
Concept for the undead beast. This enemy would become the mini-boss Drax in the graveyard / catacomb portion of the story, but was also intended to be used as a regular undead creature as well.
An unused concept for Shell 05. This tank would have most likely appeared as a mini-boss. It might have started as a design for Ankara, before being switched to Jinam. Parts of the concept might have been used in ruined battlefield debris.
An unused concept for Shell 05. This tank would have most likely appeared as a mini-boss. It might have started as a design for Ankara, before being switched to Jinam. Parts of the concept might have been used in ruined battlefield debris.
Concept for the World Bazaar on Shell 04.
Concept for the World Bazaar on Shell 04.
General Campbell’s flagship, from Ankara on Shell 05. Flying warships were something we discussed a lot. It was important to ensure that each world shell had a distinct style of ship and warfare. While the Chosen dominated from the sky, the factions of world shell 05 needed ways to deploy ground troops.
General Campbell’s flagship, from Ankara on Shell 05. Flying warships were something we discussed a lot. It was important to ensure that each world shell had a distinct style of ship and warfare. While the Chosen dominated from the sky, the factions of world shell 05 needed ways to deploy ground troops.
Concept of Marduk’s temple. I wanted something that would be ancient, but feel as if it was powered by advanced core-powered technology.
Concept of Marduk’s temple. I wanted something that would be ancient, but feel as if it was powered by advanced core-powered technology.
Kye’s design for Lobo’s ship. This was intended to look like Jinam’s signature technology, and even mirror Lobo’s head a bit. In retrospect, I am shocked that we never gave this vessel an actual name…
Kye’s design for Lobo’s ship. This was intended to look like Jinam’s signature technology, and even mirror Lobo’s head a bit. In retrospect, I am shocked that we never gave this vessel an actual name…
A design concept for the ship engine boss. The idea was that each switch was a target that could be attacked. Flipping them would eventually expose the weak points of the boss.
A design concept for the ship engine boss. The idea was that each switch was a target that could be attacked. Flipping them would eventually expose the weak points of the boss.
Concept for the swarm master, an outlaw enemy that could spawn bees from the hive on his back.
Concept for the swarm master, an outlaw enemy that could spawn bees from the hive on his back.
Concept for the tentacle brain, a bio-organic boss weapon. It controlled of the doomsday tentacles that spread through a continent to alter its gravitational orbit.
Concept for the tentacle brain, a bio-organic boss weapon. It controlled of the doomsday tentacles that spread through a continent to alter its gravitational orbit.
As brainstorming petered off I worked to finish the game design document and dove into writing the script. Waiting for the green light from VNM was extra time I did not waste: I was able to do extra research into real-world history and mythology, pump up the themes of the story, and polish the dialogue as I worked towards a final draft.
Meanwhile, we were modeling our characters while hammering out how to build our level environments. Our older games (such as Super Nintendo side scrollers) had used repeating 8×8 pixel “stamps” or “tiles” to create their environments. While that was industry standard and feasible for Septerra Core, it wasn’t desirable. Certainly there were plenty of games out there that used this method, but they looked flat, and usually felt less than organic than what we wanted.
Again, our old tools reared their blocky 320×240 heads—our old background layout tools just weren’t going to cut it. We explored using Photoshop to put our levels together in layers, but Photoshop in 1996 was pretty basic, and, like Toon Boom, it certainly wasn’t built to manage restrictive palettes and reusable game assets. We struggled to find a way to use it as a solution.
Then a talented young artist named Chip saw what we were trying to do and wanted to help.
“(I) Wasn’t sure of the concept of reusing tiles, (we were) having a difficult time,” lead artist Kathy remembers. “Then Chip , who wasn’t officially on the project, produced an incredibly beautiful rendering of one of the levels. I was dazzled, elated, relieved. I knew then that we’d be okay.”
Chip’s environment concept would go on to define the style and feel of much of Septerra Core‘s environments (here seen as part of an early demo from a bit later, in ’97). When the game finallylaunched in ’99 I was delighted by a conversation with a developer from another isometric style game, Jagged Alliance 2. While his game’s art was great too, he was flabbergasted by our environments, remarking that both of our games were isometric, yet somehow Septerra Core sure didn’t look it!
Chip had given us a goal to shoot for, and much like with our animation tools, we didn’t want to settle on what other development houses were doing with off-the-shelf software. We wanted to push the limits of the isometric view, so we set out to create a new, robust tool. Rather than rigid stamps, our environments would be built with thousands of unique, organic objects laid out in dozens of vertical layers, with complex interactive data added to create isometric worlds that would feel more organic, natural, and lived in than the games of the previous generation.
“The level layout with being able to go below or behind the scenery was quite advanced for the time.” Mark Manyen remembers. “Much of the code ran on the PSX (now called The PSOne or somesuch), now that was cool.”
A sample of some of the original background tests put together as we created our first internal demo, and later the game. Select any image to enlarge, and select again to learn more.
The world bazaar in the editing tool, Terrabuilder. Here the artist could construct their layers out of parts (floor sections, ramps, building walls, and other objects, such as the trash bins and debris scattered around.)
The world bazaar in the editing tool, Terrabuilder. Here the artist could construct their layers out of parts (floor sections, ramps, building walls, and other objects, such as the trash bins and debris scattered around.)
Early test layout in the tool. The artists had dummy character images they could place to keep a sense of scale as they laid out their environments.
Early test layout in the tool. The artists had dummy character images they could place to keep a sense of scale as they laid out their environments.
An early mock up of a forest environment, with characters added. Made up of layers of organic objects (floor tiles, rocks, and various foliage), we used these tests as bench marks to figure out what our background editing software would need to be capable of.
An early mock up of a forest environment, with characters added. Made up of layers of organic objects (floor tiles, rocks, and various foliage), we used these tests as bench marks to figure out what our background editing software would need to be capable of.
An early example of a multi-tiered environment. The characters could cross this bridge, turn and run down a ramp to the canyon floor, then run under the bridge. This was achieved by the layout artist utilizing layers of flat vector shapes to define the different areas and elevations of surfaces in the environments.
An early example of a multi-tiered environment. The characters could cross this bridge, turn and run down a ramp to the canyon floor, then run under the bridge. This was achieved by the layout artist utilizing layers of flat vector shapes to define the different areas and elevations of surfaces in the environments.
The bridge from Shell 05 being constructed in Terrabuilder. The guide lines of the editor are visible, and help the layout artist align all objects and avoid gaps in the floors.
The bridge from Shell 05 being constructed in Terrabuilder. The guide lines of the editor are visible, and help the layout artist align all objects and avoid gaps in the floors.
An actual level in the game engine with characters running around. A level like this would be made up not of rigid tiles, but hundreds of objects (each barrel, each wall section, each floor section, each pot) arranged in a pseudo 3D space.
An actual level in the game engine with characters running around. A level like this would be made up not of rigid tiles, but hundreds of objects (each barrel, each wall section, each floor section, each pot) arranged in a pseudo 3D space.
The lab being laid out. Guides have been turned on so the artist can understand the perspective of the floor and use it to align individual objects (the floor itself is made up of the diamond shapes, while the walls and other objects are then aligned on top).
The lab being laid out. Guides have been turned on so the artist can understand the perspective of the floor and use it to align individual objects (the floor itself is made up of the diamond shapes, while the walls and other objects are then aligned on top).
Another early render, this time of the Junker’s Bar. Many of the building interiors, such as shops, were much smaller in scope than other sprawling levels. As such, we could use the memory allotted for a level in RAM to add much more detail, including large rendered objects, such as the bar, lots of unique chairs, or the detailed iron girders.
Another early render, this time of the Junker’s Bar. Many of the building interiors, such as shops, were much smaller in scope than other sprawling levels. As such, we could use the memory allotted for a level in RAM to add much more detail, including large rendered objects, such as the bar, lots of unique chairs, or the detailed iron girders.
An early test of Junk City. The characters were added to this mock up of the level to show how they would look standing on a ramp between levels of the city.
An early test of Junk City. The characters were added to this mock up of the level to show how they would look standing on a ramp between levels of the city.
A mock up of a “side view” environment. We wanted to give the game a few “cinematic” locations that were not locked to the top-down three quarter view. To make this work, we would eventually limit the characters’ animation choices—forcing them to only use their left and right walking animations. This made the game appear as more of a “side scroller” for the duration of these environments.
A mock up of a “side view” environment. We wanted to give the game a few “cinematic” locations that were not locked to the top-down three quarter view. To make this work, we would eventually limit the characters’ animation choices—forcing them to only use their left and right walking animations. This made the game appear as more of a “side scroller” for the duration of these environments.
Running in Place
I think we actually spent about a year in this loop—designing the game, building our platforms, innovating technology, and going back to the suits for a greenlight, eager to start actual development with a full budget and expanded team.
But, invariably, they would only say, “We love what you’re doing—just keep going for now.”
In hindsight , a year of pre-production, while unprecedented in the stone-age past of 1996, was a massive blessing. Previous games that I had worked on had been completed in as little as 9-12 months, from concept to final gold master. That never really left much time to plan at the beginning or polish at the end. But with Septerra Core we had time to really get a clear understanding of what we were building, trim fat, focus on core gameplay, revise our scope, and iron out technical issues. This gave us a solid plan—we knew when development finally began in earnest we weren’t going to be flying by the seat of our pants.
Still, we didn’t understand why the company wasn’t giving us the go ahead. They must have liked what they were seeing or they wouldn’t have continued to fund the game. In fact, there was a push at VNM to explore merchandising for SepterraCore—they even started talking about publishing a novelization of the story. Several writers within the company auditioned for the job. Dave Martina from the IT department had a writing background and decided to throw his hat in the ring. Rather than give us unrelated writing samples, he actually borrowed some of my script and used it to write a sample chapter of a Septerra Core novel (and I liked one of the lines he gave to Doskias so much that it wound up in the game!)
Do you want to read it? Of course you do! Click below to download in PDF format.
But in the end there was only so much pre-prep we could do—sooner or later we needed to jump into full-fledged development. Certainly the company hadn’t spent so much cash on a game they weren’t going to greenlight, right?
And while we were treadmilling along, games like Diablo and Final Fantasy 7 were being announced, games which were definitely going to fill that “RPG Desert” the marketing department had wanted Septerra Core to fill. If we wanted to get to market first we needed to get into full development, and do it fast!
But we soon learned why the suits were hemming and hawing about giving us the full go-ahead:
Viacom New Media only had a few more months of life left, and then we’d all be out of a job.
